Dramatic increase in water use efficiency with cumulative forest disturbance at the large forested watershed scale

Table 3 Gradient Boosting Machine model results for selected variables

Variable	R²	RMSE	Relative importance of independent predictor variables: forest disturbance and climate
Variable	R²	RMSE	CECA	Tmax	Tmin	P	PET
Q	0.80	8	16.8	11.6	30.5	25.7	15.5
Runoff ratio	0.86	0.03	0.0	11.3	30.3	41.6	16.8
ET	0.90	29	29.1	0.0	0.0	70.9	0.0
WUE	0.93	0.04	44.4	0.0	0.0	55.6	0.0

Where each predictor variable’s relative importance ranges from 0 to 100, CECA cumulative clear-cut area, Tmean annual average daily temperature, Tmin annual minimum daily temperature, Tmax annual maximum daily temperature, P precipitation, PET potential evapotranspiration, ET evapotranspiration, Q mean annual streamflow, WUE water use efficiency, and RMSE root mean square error

ISSN: 1750-0680